Solving the bracketing paradox - German Grammar Group FU Berlin

More documents

Recommendations

Info

ST. MÜLLER however, proven to be useful to distinguish the verbal complement from other complements (Chung 1993; Rentier 1994; Müller 1997; Kathol 1998). For the purpose of representing the information about complements that form a predicate complex with their head, the feature VCOMP is introduced. Its value is a list that contains a synsem object if the verb selects for a dependent to form a complex with, and the empty list otherwise. The description in (17) shows the CAT value for the stem entry of the future tense auxiliary werden. 7 (17) werden ‘will’ future tense auxiliary ⎡ ⎤ HEAD verb ⎢ ⎥ ⎣SUBCAT 1 ⎦ 〈〉 VCOMP V[bse, SUBCAT 1 , VCOMP 〈〉] Werden selects a verb or a verbal complex via VCOMP. All arguments of this verbal complex (1) are raised. 8 The instantiations of the list 1 may be the empty list. Werden does not assign thematic roles to dependents of the embedded verb. Therefore no reference to elements possibly contained in SUBCAT is necessary. Lexical entries for the perfect auxiliaries (haben/sein) are completely analogous to (17) except for the verb form of the selected verbal complex. As Hinrichs & Nakazawa (1994) have shown, it is reasonable to assume a schema that licenses the verbal complex in addition to the head complement schema. I suggest the following schema, which licenses head cluster structures. Schema 1 (cluster schema) head-cluster-structurep ⎡ ⎤ SYNSEM|LOC|CAT|VCOMP 1 ⎢ ⎣HEAD-DTR|SYNSEM|LOC|CAT|VCOMP 1 ⊕ 〈〉 ⎥ 2 ⎦ 〈〉 NON-HEAD-DTRS [SYNSEM 2 ] A head is combined with its verbal complement (2). The remainder of the VCOMP list (1) is passed up to the mother node. In our example, 1 will be the empty list. The specification of the VCOMP value of the verbal complement of verbs like werden ‘will’ as the empty list ensures that the verbal complex that [7] For explanatory purposes, I assume that both subjects and complements are represented on the SUBCAT list. The issues discussed in this paper are orthogonal to the representation of the subject. Representations like the ones suggested in Pollard & Sag (1994: chapter 9), Kiss (1995), or Pollard (1996) are also compatible with the analysis. [8] The lexical entry as given in (17) admits multiple analyses of sentences containing this auxiliary since it is not specified that the verb that is embedded has to be lexical. Since it is not relevant to the rest of this paper, I have omitted the necessary specifications in lexical entries and in schema 1. 286
BRACKETING PARADOX AND GERMAN PARTICLE VERBS ⎡ ⎤ HEAD 1 ⎣SUBCAT 2 ⎦ VCOMP 〈〉 CL H 4 ⎡ ⎡ [ ] ⎤⎤ VFORM bse HEAD verb LOC 〈〉 ⎢ ⎢SUBCAT 2 NP[str], NP[ldat] ⎥ ⎣ ⎣ ⎦⎥ ⎦ VCOMP 〈〉 ⎡ [ ] ⎤ VFORM fin HEAD 1 verb ⎢ ⎥ ⎣SUBCAT 2 ⎦ VCOMP 〈 4 〉 helfen wird Figure 4 Analysis of the verbal complex daß er dem Mann helfen wird ‘that he will help the man’ is embedded under werden is complete, i.e., sentences like (18b), where the verb under haben ‘to have’ is missing, are ruled out. (18) (a) daß er dem Mann [[geholfen haben] wird] that he the man helped have will ‘that he will have helped the man’ (b) *daß er dem Mann haben wird that he the man have will How the analysis of the verbal complex in (16) works in detail is shown in figure 4. 9 The future auxiliary wird embeds the infinitive helfen (a verb with VFORM bse). Since no complements get saturated in head-cluster-structures, the SUBCAT list of the head is identical to the SUBCAT list of the mother. Because of this constraint, the SUBCAT list of helfen wird ‘help will’ is identical to the SUBCAT list of wird. Helfen wird is a complex head that is combined with its arguments in normal head complement structures. After this brief exposition of the analysis of German verbal complexes, I now sketch the analysis of particle verb combinations. [9] The str in the lexical entry for helfen stands for structural case. Structural case is assigned by a Case Principle that is similar to the one suggested by Yip, Maling & Jackendoff (1987): in verbal environments the first NP in the SUBCAT list that has structural case is realized as nominative and the second NP with structural case is realized as accusative. For a formalization of the Case Principle see Meurers (1999, 2000: section 10.4.1.4.), Przepiórkowski (1999). 287
Page 1 and 2: J. Linguistics 39 (2003), 275-325.
Page 3 and 4: BRACKETING PARADOX AND GERMAN PARTI
Page 11: BRACKETING PARADOX AND GERMAN PARTI
Page 51: BRACKETING PARADOX AND GERMAN PARTI

Solving the bracketing paradox - German Grammar Group FU Berlin

Create successful ePaper yourself

Delete template?

Save as template?